Objectives: Studies of saccadic eye movement impairment in Alzheimer’s disease (AD) have largely focused on simple reflexive eye movements and the antisaccade task. The effects of manipulating stimulus timing have been little studied. Methods: Fourteen patients with mild to severe AD and 11 age-matched controls were studied on the antisaccade task, while latencies on simultaneous, gap and predictable tasks were recorded from 11 patients and 11 controls. Dementia severity was assessed with the Mini-Mental State Examination. Results: As a group, patients’ latencies were significantly higher and more variable on the simultaneous and gap tasks. Predictable task performance was similar in mean but significantly more variable. Grossly anticipatory responses by patients were common on the predictable, simultaneous and gap tasks. Exclusion of these from subject means revealed that AD patients, when making target-driven saccades, demonstrated a gap effect of similar magnitude to normal subjects. Patients made significantly fewer correct antisaccades and significantly more reflexive errors not followed by a corrective antisaccade than did controls. Conclusions: The frequent presence of grossly anticipatory saccades may reflect dysfunction of fixation mechanisms possibly involving projections from frontal lobe to superior colliculus. The less frequently seen, marked prolongation of latency may reflect changes in posterior parietal mechanisms mediating reflexive saccade generation. The presence of the gap effect demonstrates a continued ability to benefit from externally controlled stimulus disengagement. Patients’ ability to make appropriately timed saccades to targets of known locations was particularly impaired, but the target sequence itself was at least grossly correctly learned. Larger studies may be able to identify clinically distinct populations of AD patients.